Maximal multiplier operators in \(L^{p(\cdot)}(\mathbb{R}^n)\) spaces (Q281946)

The authors study when the Fourier transform of a compactly supported measure becomes a multiplier on variable Lebesgue spaces. Namely, they extend \textit{J. L. Rubio de Francia}'s theorem [Duke Math. J. 53, 395--404 (1986; Zbl 0612.42008)] and, more concretely, its modified proof by \textit{A. Iosevich} and \textit{E. Sawyer} [Adv. Math. 132, No. 1, 46--119 (1997; Zbl 0921.42015)] to the setting of variable Lebesgue spaces \(L^{p(\cdot)}(\mathbb R^n)\). Their main result establishes that if \(m(\xi)\) is the Fourier transform of a compactly supported locally \(\beta\)-dimensional Borel measure \(\sigma\) for some \(0\leq \beta\leq n\) satisfying that \[ \max\{\int_1^2 |\nabla m(t\xi)|^2dt, \int_1^2 |m(t\xi)|^2\} \leq C (1+|\xi|)^{2\alpha} \] for some constants \(C>0\) and \(\alpha>1/2\), where \(p(\cdot)\) is a measurable function with values in \([1,\infty]\) such that the Hardy-Littlewood maximal function is bounded in \(L^{q(\cdot)}(\mathbb R^n)\) for \(q(\cdot)=\frac{2\theta p(\cdot)}{1-(1-\theta)p(\cdot)}\) for some \(0<\theta< \frac{2\alpha-1}{2\alpha-1+2(n-\beta)}\), then the maximal operator \(\mathcal M_m\) is bounded in \(L^{p(\cdot)}(\mathbb R^n)\), where \(\mathcal M_m f(x)=\sup_{t>0}|\int_S f(x-ty)d\sigma(y)|\), \(m(\xi)=\hat\sigma(\xi)\) and \(\sigma\) is supported on the compact set \(S\). Further consequences are achieved when the variable exponent \(p(\cdot)\) is globally log-Hölder continuous and certain restrictions on the values \(p_{-}=\text{essinf} \, p(x)\) and \(p_+=\text{esssup} \, p(x)\) are imposed.